ABSTRACT

With the explosive growth of Big Data and Internet of Things, and the ever-growing demand for high-performance computing, mobility, security, and high-fidelity experiences, there is an increasing need for high-bandwidth and low-loss interconnect technologies that enable efficient data transfer across the chip. The optical fiber and its significant capacity for high-bandwidth data transfer are well-known in the telecommunications industry, and widely exploited in their long-distance network of interconnects. Scaling down optical interconnects to fit in nano-scale silicon chips would clearly be advantageous for achieving high-bandwidth data transfer on-chip -- however, a major hurdle is the dearth of accurate and efficient computable stochastic electromagnetic models for simulation of optical interconnect structures comprised of multiple tightly-coupled nano-scale silicon-on-insulator (SOI) wave-guides that carry information signals (i.e., TeraHertz electromagnetic waves) across on-chip transmitters and receivers. The overall goal of this proposal is to develop algorithms for a software tool to perform the design, analysis, and optimization of 3-dimensional (3-D) nano-scale optical interconnects based on SOI wave-guides exhibiting random surface roughness. The overall educational components of this project are to leverage the developed models and software to: (1) develop new graduate courses, (2) develop interactive learning objects and lab-based activities for undergraduate courses, and (3) stimulate undergraduate students' interest in science, and recruit and mentor diverse groups of students including women and minority groups.

The project team will develop the Optical Interconnect Designer Tool (OIDT) software, of which the input is comprised of: (1) the specified 3-D geometry representing the physical description of the multi-port optical interconnect system, (2) the random distribution for surface roughness of SOI wave-guide, and (3) the wave-guide material's electrical properties. The OIDT will autonomously synthesize two types of electrical models: (1) network scattering parameters for design optimization of the interconnect, in the frequency-domain, and (2) stable, passive, and causal SPICE equivalent circuit models for timing analysis and signal/power integrity analysis of the passive interconnect, integrated with active non-linear drivers and components, in the time-domain. The proposed Python-based software package may be used stand-alone, or integrated into existing computer aided design (CAD) tools and design-flows to facilitate design automation, and research and development of advanced microelectronics for a variety of applications including computing, communications, energy, security, sensing, health, etc. The numerical program will be hosted on GitHub as an open-source project, to facilitate and promote (inter)national optical device research. The education component expands public's scientific literacy.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Please report errors in award information by writing to: awardsearch@nsf.gov.